Create app.py

app.py

@@ -0,0 +1,113 @@
+import gradio as gr
+import numpy as np
+from PIL import Image
+import requests
+from pathlib import Path
+import torch, numpy as np, pandas as pd
+from fastai.data.transforms import RandomSplitter
+from torch import tensor
+
+def car_purchase():
+    np.set_printoptions(linewidth=140)
+    torch.set_printoptions(linewidth=140, sci_mode=False, edgeitems=7)
+    pd.set_option('display.width', 140)
+    
+    path = Path('/dataset')
+    
+    df = pd.read_csv(path/'Car_Purchasing_Data.csv')
+    
+    df = df.drop(columns=['Customer Name', 'Customer e-mail', 'Country'])
+    
+    gender = sorted(df.Gender.unique())
+    
+    df.rename(columns={'Gender': 'gender',
+                       'Age': 'age',
+                       'Annual Salary': 'annual_salary',
+                       'Credit Card Debt': 'credit_card_debt',
+                       'Net Worth': 'net_worth',
+                       'Car Purchase Amount': 'car_purchase_amount'},
+              inplace=True)
+    
+    t_dep = tensor(df.car_purchase_amount)
+    
+    indep_cols = ['gender', 'age', 'annual_salary', 'credit_card_debt', 'car_purchase_amount']
+    t_indep = tensor(df[indep_cols].values, dtype=torch.float)
+    
+    
+    torch.manual_seed(390)
+    
+    n_coeff = t_indep.shape[1]
+    coeffs = torch.rand(n_coeff)-0.5
+    
+    vals, indices = t_indep.max(dim=0)
+    vals_d, indices_d = t_dep.max(dim=0)
+    t_indep = t_indep / vals
+    t_dep = t_dep / vals_d
+    
+    def calc_preds(coeffs, indeps): 
+      return (indeps*coeffs).sum(axis=1)
+    
+    def calc_loss(coeffs, indeps, deps): 
+      return torch.abs(calc_preds(coeffs, indeps)-deps).mean()
+    
+    coeffs.requires_grad_()
+    
+    loss = calc_loss(coeffs, t_indep, t_dep).to(dtype=torch.float32)
+    
+    loss.backward()
+    
+    coeffs.grad
+    
+    with torch.no_grad():
+        coeffs.sub_(coeffs.grad * 0.1)
+        coeffs.grad.zero_()
+        print(calc_loss(coeffs, t_indep, t_dep).to(dtype=torch.float32))
+    
+    trn_split,val_split=RandomSplitter(seed=42)(df)
+    
+    trn_indep,val_indep = t_indep[trn_split],t_indep[val_split]
+    trn_dep,val_dep = t_dep[trn_split],t_dep[val_split]
+    len(trn_indep),len(val_indep)
+    
+    def update_coeffs(coeffs, lr):
+        coeffs.sub_(coeffs.grad * lr)
+        coeffs.grad.zero_()
+    
+    def one_epoch(coeffs, lr):
+        loss = calc_loss(coeffs, trn_indep, trn_dep)
+        loss.backward()
+        with torch.no_grad(): update_coeffs(coeffs, lr)
+        print(f"{loss:.3f}", end="; ")
+    
+    def init_coeffs(): 
+      return (torch.rand(n_coeff)-0.5).requires_grad_()
+    
+    def train_model(epochs, lr):
+        torch.manual_seed(442)
+        coeffs = init_coeffs()
+        for i in range(epochs): one_epoch(coeffs, lr=lr)
+        return coeffs
+    
+    coeffs = train_model(6, lr=0.2)
+    
+    def show_coeffs(): 
+      return dict(zip(indep_cols, coeffs.requires_grad_(False)))
+    
+    show_coeffs()
+
+demo = gr.Interface(
+    fn=car_purchase,
+    title="Car Purchase Analytics",
+    description="Experiment with the features to predict car purchase amount.",
+    allow_flagging="never",
+    inputs=[
+        gr.inputs.Number(default=30.0, label="Age"),
+        gr.inputs.Radio(['female', 'male'], label="Sex"),
+
+        gr.inputs.Number(default=1.0, label="Annual Salary"),
+        gr.inputs.Number(default=1.0, label="Credit Card Debt"),
+        gr.inputs.Number(default=1.0, label="Credit Card Debt"),
+    ],
+    outputs=gr.Image(type="pil"))
+
+demo.launch()